Cross device user identification
Data from browser requests is added to a vector. If explicit identification information (username, cookie data, etc.) is present, the vector is associated with a pre-existing user record, which is then updated. If not, candidate user records may be identified according to correspondence with values in the vector. This may include comparing hashes of one or more values to identify similarities. Candidate vectors may be eliminated by identifying inconsistency in OS, device, and browser information. Probability assigned to each candidate vector may be adjusted, e.g., reduced, in response to inconsistency in other data relating to a browser, device, or of a more global nature, e.g. time zone, user attributes, etc. Records associated with different devices may be associated with one another by evaluating hashes of data values submitted by a user on the different devices. Where the hash values of two records intersect, they may be merged with one another.
1. A method for data collection comprising:
generating, by a computer system, a plurality of records for a plurality of devices, wherein each record is associated with one device of the plurality of devices, by for each device of a plurality of devices,
receiving, by the computer system from the each device, hashes of data values entered on the each device into input fields of one of a plurality of unrelated third party websites, the hashes not being labeled with labels indicating a type of data represented by the hashes; and
storing, by the computer system, the hashes received from the each device in the record associated with the each device; and
identifying, by the computer system, first and second devices of the plurality of devices that correspond to a same user by:
identifying matching hashes of the hashes stored in the record associated with the first device that is also stored in the record associated with the second device;
determining a probability for the first device and the second device by, for each hash of the matching hashes, augmenting the probability by an amount that corresponds to infrequency of occurrence of the each hash in the records associated with the plurality of devices;
determining that the probability meets a threshold certainty; and
in response to determining that the probability meets the threshold certainty, determining that the first device and the second device correspond to the same user.
2. The method of claim 1 , further comprising merging, by the computer system, user history data received from the first and second devices into a single user record.
3. The method of claim 2 , wherein the user history data includes browsing activities of the first and second devices.
4. The method of claim 3 , further comprising:
selecting, by the computer system, content for transmission to the first device based on the browsing activities of the first and second devices; and
transmitting, by the computer system, the content to the first device.
5. The method of claim 1 , wherein the hash function is a lossy function.
6. The method of claim 1 , wherein the data values include at least one of email addresses, phone numbers, and street addresses.
7. The method of claim 1 , wherein the data values include credit card numbers.
8. A system for data collection comprising one or more processing devices and one or more memory devices operably coupled to the one or more processing devices, the one or more memory devices storing executable code that, when executed, causes the one or more processing devices to:
generate a plurality of records, each record of the plurality of records being associated with one device of a plurality of devices, by for each device of a plurality of devices:
receive, from the each device, hashes of data values entered into input fields of one of a plurality of unrelated third party websites, the hashes not being labeled with data labels indicating a type of data represented by the hashes; and
store the hashes in the record associated with the each device; and
if there are matching hashes in the record associated with a first device of the plurality of devices that are included in the hashes in the record associated with a second device of the plurality of devices:
determine a probability associated with the first device and the second device by, for each hash of the matching hashes, augment the probability by an amount that decreases with frequency of occurrence of the each hash in the plurality of records associated with the plurality of devices; and
if the probability meets a threshold certainty, determine that the first device and the second device are associated with a same user.
9. The system of claim 8 , wherein the executable code, when executed, further causes the one or more processing devices to:
if the probability meets the threshold certainty, merge user history data received from the first and second devices into a single user record.
10. The system of claim 9 , wherein the user history data includes browsing activities of the first and second devices.
11. The system of claim 10 , wherein the executable code, when executed, further causes the one or more processing devices to:
select content for transmission to the first device based on the browsing activities of the first and second devices; and
transmit the content to the first device.
12. The system of claim 8 , wherein the hash function is a lossy function.
13. The system of claim 8 , wherein the data values include at least one of email addresses, phone numbers, and street addresses.
14. The system of claim 8 , wherein the data values include credit card numbers.